OpenSearch has traveled a considerable distance since Amazon Web Services launched it in January 2021, following Elastic's shift of Elasticsearch and Kibana away from Apache 2.0 licensing toward proprietary and source-available models. Elastic justified the change by citing concerns that major cloud providers like AWS were offering Elasticsearch as a managed service without contributing back to the project, and also raised trademark violation allegations against AWS. Though Elastic later reintroduced Elasticsearch and Kibana to open source with AGPL licensing options and reconciled with AWS, OpenSearch has maintained its separate trajectory as an independent open source initiative—though one whose origins remain tied to the technology giant that created it.

Reducing the AWS dependency

A pivotal turning point occurred in September 2024, when AWS transferred OpenSearch to the Linux Foundation and established the OpenSearch Software Foundation, moving governance to a "vendor-neutral" model. This transition aimed to reassure other organizations that participation in OpenSearch would not leave AWS in control of the project's direction.

Two years following the governance shift, OpenSearch demonstrates expanding momentum independent of its founding sponsor. Data shared with The New Stack by the OpenSearch Software Foundation reveal that AWS's proportion of contribution activities within the project's repositories has declined from 91% during the initial year to 72% in 2025–26. In the core OpenSearch repository, Amazon-affiliated maintainers represented 95% of the total in April 2022, a figure that has decreased to 71% currently.

The trend manifests in deployment patterns as well. According to the 2026 Open Data Infrastructure Report, released recently by the Linux Foundation and the OpenSearch Software Foundation, self-managed deployments have expanded more rapidly than Amazon's managed OpenSearch Service beginning in 2024. Self-managed cloud deployments now account for 60% of surveyed OpenSearch users, up from 37% in 2024, whereas Amazon's managed service usage increased more gradually, rising from 44% to 48%.

AWS's historical connection to OpenSearch continues to present challenges for certain potential adopters. The report indicates that roughly one-quarter (24%) of 71 surveyed organizations that have not adopted OpenSearch identify concerns regarding vendor neutrality and the project's historical relationship with AWS as factors in their decision.

These metrics highlight a fundamental obstacle that governance restructuring alone cannot overcome: persuading prospective users and contributors that AWS no longer primarily determines OpenSearch's trajectory.

'A watershed moment' for OpenSearch

Bianca Lewis, executive director of the OpenSearch Software Foundation, discussed the project's progression beyond its Elasticsearch roots and perceptions of lingering AWS influence during a recent conversation at Open Source Summit Europe in Prague.

Lewis contends that the project's legal autonomy, secured over two years ago, establishes a critical foundation for cultivating confidence.

When they donated it to the Linux Foundation, that was a watershed moment.

Bianca Lewis, executive director of the OpenSearch Software Foundation

Lewis elaborates that the transfer meant AWS could no longer shut down OpenSearch or alter its licensing terms, having surrendered ownership of the application. She also frames the decision as commercially rational for AWS, whose cloud revenue derives from computing resources powering applications rather than from owning the applications themselves.

AWS don't make money from applications. AWS make money from compute. So operating, growing OpenSearch long term was never in AWS's plan.

Bianca Lewis

Nevertheless, questions persist regarding AWS's continuing influence. The company still represents nearly three-quarters of documented contribution activities across OpenSearch repositories. Though AWS occupies only four of 15 voting positions on the Technical Steering Committee (TSC), an AWS employee, Andrew Ross, was reelected committee chair in October.

The AI opportunity for OpenSearch

OpenSearch itself has undergone substantial transformation since the initial fork, and Lewis acknowledges the striking resemblance to Elasticsearch in its early iterations.

"It's undeniable that the early versions of OpenSearch was Elasticsearch with a different name," she states. "They had the same plugins. Even a lot of the documentation of OpenSearch was Elasticsearch documentation. [But] when it got donated to the Linux Foundation a couple years back, it had already well and truly diverged from its original mother project."

Currently, OpenSearch positions itself as foundational infrastructure for applications requiring search, analysis, and data retrieval capabilities, including those leveraging large language models (LLMs). Features such as vector and hybrid search functionality align with retrieval-augmented generation (RAG) use cases, where AI systems pull information from external repositories to enhance model outputs.

OpenSearch 3.9, released in September, expands these capabilities with enhancements targeting developers building AI applications, alongside improvements to performance, monitoring, and alerting functions.

Lewis sees potential for OpenSearch to become a cornerstone of enterprise AI infrastructure. "If you decide that you wanted your AI infrastructure on an open platform, I think [OpenSearch] today is just the default choice," she remarks.

Winning the enterprise

Strategic goals differ from the practical work of establishing sustained enterprise adoption. The Foundation's report shows that many OpenSearch deployments remain in exploratory phases, with organizations still assessing the platform or deploying it for replaceable use cases.

Lewis attributes this partly to the extended timelines companies require to transition away from entrenched commercial solutions, particularly when locked into long-term licensing arrangements. "They might have a three-year contract for Splunk, a license for five years with Elastic," she explains. "So they're not just chucking that out."

Organizations may therefore be evaluating OpenSearch while continuing to pay for existing products. However, no certainty exists that such evaluations will translate into permanent commitments.

Lewis identifies one potential driver for eventual migration: the capacity to switch technology providers without requiring application redesign.

What if I get locked in? And if I get locked in and I want to leave, what does that look like? Do I have to rearchitect all my applications?

Bianca Lewis

This concern encompasses more than licensing expenses. Enterprises must weigh the engineering costs of vendor transitions, data migration, and long-term application maintenance.

The discussion ultimately returns to OpenSearch's independence. If organizations select the platform partly to escape vendor lock-in, the project's capacity to build and sustain a contributor base outside AWS becomes increasingly vital.

When asked about her vision for OpenSearch in five years, Lewis emphasizes vendor influence and her goal for a project free from single-company dominance. "To have a truly sustainable, fast, innovating platform, which is not dominated or heavily influenced by any single vendor in the world," she says. "Then, you've got a stable platform."